Skip to content

Security: TratokToken/Tara-AI

Security

SECURITY.md

Security policy

Reporting a vulnerability

If you find a security issue in the Tara service, the SDKs in this repo, or the documented behaviour:

Don't open a public GitHub issue.

Email platform@tratok.net with subject Tara security report and include:

  1. A description of the issue.
  2. Steps to reproduce (curl is fine).
  3. Impact — what an attacker could do.
  4. Any relevant request_ids.
  5. Your contact for follow-up.

We'll acknowledge within 48 hours and aim to triage within 5 business days.

In scope

  • The API at tara.tratok.com/api/v1/* — auth bypass, privilege escalation, data leakage.
  • The web UI at tara.tratok.com — XSS, CSRF, account takeover.
  • The SDKs and example code in this repository — credential leakage, unsafe defaults.
  • The OpenAPI spec — if it documents an unsafe pattern.

Out of scope

  • Findings from automated scanners with no proof of impact. Bring a working PoC.
  • Best-practice complaints with no exploit (e.g., "you don't pin TLS"). Send a hardening suggestion instead.
  • DoS / volumetric attacks. Rate limiting is a known mitigation.
  • Self-XSS in the chat UI. It's anonymous, single-tab, no persistence — there's nothing to steal.

Coordinated disclosure

We follow a standard 90-day disclosure window:

  • Day 0: report received.
  • Day 0–14: triage, ack with severity.
  • Day 14–80: fix and deploy.
  • Day 80–90: coordinate public disclosure with reporter.

Critical issues get faster turnaround.

Bounties

No formal bounty programme yet. We do publicly credit reporters in the CHANGELOG and on tara.tratok.com. For high-impact reports we may send Tratok merchandise or platform credits.


Thanks for taking the time to report responsibly.

There aren't any published security advisories